Repolode
Submit
rapidsai

rapidsai/jupyterlab-nvdashboard

A JupyterLab extension that adds dashboards for monitoring GPU compute, memory, PCIe and NVLink usage inside notebooks.

View on GitHub
Stars
681
Forks
89
Open issues
34
Last commit
11 days ago

What is it?

JupyterLab NVdashboard is an extension for JupyterLab that displays GPU usage dashboards alongside the notebooks and files you already work in. It charts system hardware metrics such as GPU-compute utilization, GPU-memory consumption, PCIe throughput and NVLink throughput, so you can watch hardware activity without leaving the development environment.

The charts support brushing to select a time range for closer inspection, and tooltips on multi-chart pages align at the same timestamp for comparison. The dashboard follows JupyterLab's light and dark themes.

The extension also includes a GPU Accel menu in the notebook toolbar. It can switch on cuDF pandas and cuML accelerators for scikit-learn, UMAP and HDBSCAN code without changing the code, when the matching libraries are installed in the notebook's environment. The v4 release targets JupyterLab 4 and later, and it can be installed with conda or pip.

Key features

  • 01Displays GPU compute utilization and GPU memory consumption charts.
  • 02Shows PCIe and NVLink throughput metrics.
  • 03Lets you drag to select a time range on time series charts.
  • 04Aligns tooltips across charts at the same timestamp.
  • 05Toggles cuDF pandas and cuML accelerators from the notebook toolbar without code changes.

Use cases

  • Monitoring GPU load while training or running models in a JupyterLab notebook.
  • Checking whether pandas or scikit-learn code is actually using the GPU after enabling acceleration.
  • Comparing memory and throughput metrics over a past time window during debugging.

Quick start

  1. 1

    Install from PyPI (stable)

    pip install jupyterlab-nvdashboard
  2. 2

    Or install from conda-forge (stable)

    conda install -c conda-forge jupyterlab-nvdashboard

Things to know

  • Requires JupyterLab v4 or later, since v4 supports only JupyterLab v4+.
  • GPU accelerator controls appear only when cuDF or cuML are installed in the notebook environment.

In collections

Run and train AI on GPUsDashboards & business intelligence toolsData science courses & tools

Topics

#gpu#jupyterlab#monitoring#dashboard#cuda#nvidia#data-science

FAQ

What is jupyterlab-nvdashboard?
A JupyterLab extension that adds dashboards for monitoring GPU compute, memory, PCIe and NVLink usage inside notebooks. JupyterLab NVdashboard is an extension for JupyterLab that displays GPU usage dashboards alongside the notebooks and files you already work in. It charts system hardware metrics such as GPU-compute utilization, GPU-memory consumption, PCIe throughput and NVLink throughput, so you can watch hardware activity without leaving the development environment.
Is jupyterlab-nvdashboard free to use? What license does it have?
jupyterlab-nvdashboard is open source under the BSD-3-Clause license. Check the license terms before using it in commercial products.
Is jupyterlab-nvdashboard still maintained?
Actively maintained. The last commit to jupyterlab-nvdashboard was on Oct 27, 2026.
What can I use jupyterlab-nvdashboard for?
Monitoring GPU load while training or running models in a JupyterLab notebook. Checking whether pandas or scikit-learn code is actually using the GPU after enabling acceleration. Comparing memory and throughput metrics over a past time window during debugging.
How do I install jupyterlab-nvdashboard?
Install from PyPI (stable): pip install jupyterlab-nvdashboard Or install from conda-forge (stable): conda install -c conda-forge jupyterlab-nvdashboard
Where can I get jupyterlab-nvdashboard?
The source code and installation instructions are on GitHub: https://github.com/rapidsai/jupyterlab-nvdashboard

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects